Popular repositories Loading

1. DeepSeekV3Mini DeepSeekV3Mini Public

   Miniature implementation and training of DeepSeekV3 in order to test my understanding of LLM training pipelines and ability to implement a paper.

   Python

2. Pixie Pixie Public

   An attempt at implementing a Genie clone on the tinyworlds dataset.

   Python

3. MiniDreamer MiniDreamer Public

   An attempt at implementing a Dreamer-style agent on space invaders

   Python

4. DecoderFreeMiniDreamer DecoderFreeMiniDreamer Public

   A decoder free version of my previous dreamer agent that uses a barlow twins style self-supervised learning objective instead of a reconstruction loss.

   Python

5. Voyager Voyager Public

   A dreamer-style agent that attempts to solve the infinite horizon credit assignment problem

   Python